Decades after the development of effective therapy, TB remains one of the world's deadliest infections, with 9.7 million new cases and 1.8 million deaths in 2007. In resource-limited settings, lack of access to skilled laboratories is a bottleneck to diagnosis, particularly for MDR-TB. Patients are generally put on first-line therapy, and treatment failure is only discovered four months later. Those who fail treatment, for whatever reason, can continue to be infectious, spreading disease to others and running increasing risks of morbidity and mortality. A lab-free way to identify those who are failing treatment would bypass this problem, allowing clinicians without access to culture to make a crucial determination that affects both the patient outcome and public health safety - is the patient responding to the provided therapy? Cough frequency dynamics, which measure the changes in how many times a person coughs over the course of treatment, may be an ideal lab-free way to diagnosis treatment failures. Our preliminary data suggests that a conversion to negativity of mycobacterial culture is paralleled by a decrease in cough frequency. In this application, we propose to evaluate this more extensively. We will accomplish this through the following Specific Aims: Specific Aim 1: We will assess the correlation of cough frequency dynamics and treatment response in patients with sensitive and MDR-TB, as defined by changes in sputum smear and culture results. Specific Aim 2: We will evaluate the correlation of cough frequency dynamics with clinical severity, radiological extent of disease by chest x-ray and CT scan, and TB resistance patterns. In the resource-limited settings where TB thrives but access to appropriate technology and laboratories is limited, cough provides an ideal tool for the early identification of treatment failures without a laboratory, as well as an objective indication of patients are no longer transmitting. If the studies proposed here are successful, we would perform a larger study in which we evaluate an appropriate sample size to determine sensitivity and specificity of cough frequency as a determinant of treatment success. As in this smaller pilot study, we would examine subjects with a range of clinical severities, and subjects with both drug-sensitive and drug-resistant TB. In addition, we would examine the role of HIV infection in treatment failure and the capability of cough monitoring in HIV+ cases of pulmonary TB. We would also begin research into the development of a cough counter with appropriate characteristics for implementation in the developing world.